Dynamically Naming Expressions and Constraints in JuMP

Specific Domains Optimization (Mathematical)
,
1

In JuMP, it appears that expressions, constraints, and variables cannot be dynamically named using string interpolation (i.e., the $ symbol) directly within the identifiers of the @variable or @expression macros. Given this limitation, I am curious about the best approach to dynamically generate names for expressions and constraints within a loop.

Currently, my workaround involves defining an empty dictionary outside the loop, which is then populated with expressions and constraints as follows:

exp = Dict()
for s in 1:S
    exp[s] = @expression(model, sum(a[i] * x[i] for i in 1:n))
    # Additional code follows
end

Is this approach considered the recommended practice for naming expressions and constraints dynamically?

1 Like
2

Currently, my workaround involves defining an empty dictionary outside the loop, which is then populated with expressions and constraints as follows:

Yes, this is a good approach.

JuMP variables and expressions are just regular Julia types. They are not special. The JuMP macros provide some convenient constructors, like @expression(model, exp[s in 1:S], ...), but you don’t need to use them. Use whatever data structures are most convenient for your problem.

See also, parts of the docs like this Variables · JuMP, or tutorials like The network multi-commodity flow problem · JuMP.